Agile Brand Guide Holiday Shopping Data and Predictions

Spending More, Buying Less: Our 2026 Holiday Predictions

Why Our 2026 Holiday Sales Predictions Fall Below Consensus

Most of the major holiday forecasts are in, and they cluster around 4% growth or better. NRF expects 4.2%, EMARKETER 4.1%, Bain 4.5%, ICSC 4.3% to 4.9%, and Mastercard 5.5%. We expect 3.0% to 3.5%.

The gap comes down to what we weight. The consensus leans on a labor market that is still adding jobs and on consumers who have kept spending through three years of gloomy surveys. We’ve spent the past months compiling holiday data going back to 1992 for our Holiday Shopping Tracker, and what stands out in that longer record is different: the pre-season indicators this year are the weakest we have on file, shoppers’ own spending plans are flat to down, and most of the growth that forecasters are counting on is price, not volume.

This is our first year making predictions of our own, so some context on how we’re doing it. We’re publishing four specific, numeric calls before the season starts, and in January we’ll grade them on the tracker with the same government and industry data we use to grade everyone else. Below are the four predictions and the reasoning behind each.

Revisiting past forecasts versus the actuals

First, the field. Holiday forecasting is genuinely hard: the season is two months of spending shaped by prices, weather, the calendar, and whatever happens in the news between now and New Year’s. The forecasters we track do serious work with good data, and we use their numbers throughout the tracker. Disagreeing with the consensus is not a knock on the people who built it.

We built the tracker to see how predictions compare with outcomes over time, and that comparison will point out a few things for everyone involved (including us in the future, I am sure).

NRF has published a forecast range every year since 2017. Measured against the Census Bureau data NRF itself uses as its baseline today, actual growth has never landed inside that range for the last nine seasons. Setting aside the pandemic years, which scrambled every model, the misses have leaned in one direction: actuals came in below the forecast in six of eight seasons since 2015, by about 0.6 percentage points on average.

One caveat we want to state plainly, for fairness’ sake: we score forecasts against today’s revised Census figures, not against the preliminary numbers available when each forecast was graded at the time. Revisions move these numbers, sometimes by a lot. Last season is a good example: the January headline said holiday sales grew 4.1%, based on card-transaction data, while the Census series NRF now puts 2025 growth at 3.1%. Neither number is wrong. They measure things differently, and part of the tracker’s job is to keep them separate.

The lesson we take isn’t that forecasts are bad. It’s that the published ranges have tended to run a little ahead of what the Census data later showed, and that an honest forecast this year should account for that.

Prediction 1: Total holiday sales grow 3.0% to 3.5%

We expect holiday sales of roughly $1.0 trillion, up 3.0% to 3.5% from 2025’s $972.3 billion, using NRF’s definition (November 1 through December 31, excluding autos, gas, and restaurants, measured by Census data). That would still be a record in dollar terms, but a less steep climb than most forecasts call for.

Here’s what pushes us below the consensus:

Consumers are entering the season in a worse mood than any year in our data.

The University of Michigan’s consumer sentiment index finished October at 48.0, the lowest pre-season reading in the tracker, below even 2022, when inflation was running above 7%. The Conference Board’s confidence index tells a similar story, with an asterisk: it rebounded to 91.7 in October from September’s 84.2, the weakest reading since 2014. The Conference Board credits the October bounce largely to gas prices easing, and even with it, confidence sits below last year’s pre-season 94.6.

The budget math is tighter.

Gasoline averages $4.20 a gallon, down from about $4.43 a month ago but well above the $3.04 drivers paid a year ago, and higher than any previous pre-season reading in our data, including 2022’s $4.02. The personal saving rate fell to 3.8% in September, the lowest since November 2022, and real disposable income is barely growing.

Shoppers’ own plans are flat to down.

Circana’s survey has planned holiday spending slipping from $796 to $779 per shopper. YouGov finds 17% of Americans planning to spend less this season against 15% planning to spend more. Deloitte’s per-shopper figure is up modestly, so the surveys don’t all agree, but none of them describes a consumer planning a 4%-to-5% increase.

Two counterpoints are worth considering here, though. Sentiment has been a poor predictor of actual spending: 2022, 2023, and 2025 all had weak surveys and still produced growth between 1.9% and 4.0%. The calendar helps this year, with 28 days between Thanksgiving and New Year’s Eve, up from 27 last year. That’s why our range starts at 3.0% rather than something more dramatic. We’re calling for a season that’s a bit below what forecasters expect, not a bad one.

Prediction 2: Real growth of 0.5% to 1.5%, because this season is about price, not volume

This is the prediction we’d most like readers to sit with, because it reframes the first one.

Last season, holiday sales grew 3.1% in nominal dollars. Strip out the change in goods prices and real growth was 1.7%. Adjusted for overall inflation, holiday sales peaked in 2021 and haven’t gotten back there since: real growth was negative in 2022 and 2023 and under 1% in both 2024 and 2025.

We expect 2026 to follow the pattern. With tariffs still working their way into goods prices, retailers can post a headline number near 4% while selling roughly the same number of items as last year. NRF’s own forecast assumes core goods inflation of 2.5% this year, up from 1.6% in 2025, which tells the same story from the other direction. Our call is real growth of 0.5% to 1.5% on the core-goods deflator, below last year’s 1.7%.

The practical version: shoppers will spend more and bring home about the same. For retailers, that means the season’s results will depend more on pricing and margins than on traffic. For anyone reading January’s headlines, it means the nominal growth number alone won’t tell you whether the season was actually good.

Fair warning on this one: it depends on November and December inflation, which nobody knows yet. If goods prices cool faster than expected, real growth could land above our range even if the nominal number comes in where we think.

Prediction 3: Seasonal retail hiring falls below 360,000

Every October through December, retailers add workers for the rush, and the Bureau of Labor Statistics counts them. That count has been shrinking: 434,400 in 2021, then 373,000, 397,900, 398,300, and 362,600 last year, the fewest since the 2008 recession. For scale, the peak was 676,100 in 1999.

We expect the slide to continue: 320,000 to 360,000 seasonal hires this year, our range centering about 6% below last season.

Part of this is the sales picture above. A retailer planning for 3% nominal growth that’s mostly price doesn’t need more floor staff. Part of it is structural: more volume keeps shifting online, fulfillment keeps automating, and self-checkout and AI-assisted customer service reduce how many seasonal workers each dollar of sales requires. We’d be careful about attributing the whole decline to AI, since the trend predates the current wave of tools, but the direction is consistent.

One measurement note: NRF publishes its own seasonal hiring figures using a different method, so the two series don’t always agree (NRF counted 442,000 for 2024 against the BLS’s 398,300, and estimates 356,000 for 2025, a bit under the BLS figure). NRF’s 2026 outlook, released October 15 alongside its sales forecast, calls for 325,000 to 375,000 seasonal hires. Our range sits a notch below that, and it’s scored against the BLS series, which is the one most press coverage quotes.

Prediction 4: AI changes how people shop this season, but not yet how they buy

This is a two-part call, and it comes from the AI Shopping Index we maintain on the tracker.

Part one: AI-referred traffic to retail sites beats Adobe’s forecast again. Adobe measures visits to retail sites that arrive from generative AI tools. That traffic grew about 1,300% in the 2024 season and 693% in 2025, when Adobe had forecast 520% – the actual came in well above its own projection. For 2026, Adobe forecasts 130% growth, and the most recent monthly reading, August, ran at 127%. We expect the season to land between 130% and 200%, above Adobe’s number once holiday promotion kicks in. To be clear about what we respect here: Adobe’s team has been conservative with a metric that’s genuinely hard to project off a fast-growing base, and its day-level 2025 forecasts all came in under the actuals too. We’re betting the pattern holds one more year.

Part two: AI agents complete fewer than 3% of US holiday online orders. The agents are real now. Meta Muse, Alexa for Shopping, ChatGPT, Perplexity, Copilot, and Google’s shopping features can all research products, and several can check out on a shopper’s behalf. But the survey data shows a wide gap between using AI to shop and letting it buy.

Deloitte finds 39% of holiday shoppers calling generative AI a primary shopping tool. YouGov finds 43% of likely Black Friday shoppers would let AI find them the best price, while 4% would let it complete a purchase without asking first. Even the most generous survey wording we track – from Cordial, Product.ai, and an October study by Simon-Kucher – top out at 13% to 16% willing. There’s a trust gap underneath: AI shopping tools earn 48% consumer confidence against 69% to 80% for established channels, and 89% of consumers say they double-check AI answers.

So our call is that this season’s AI story is research and traffic, not autonomous checkout. Expect headlines about agentic shopping; the transaction share will still be small.

One honesty requirement we’ve set for ourselves: nobody currently publishes a clean measure of agent-completed holiday orders. The numbers that exist measure something broader. Salesforce, for example, tied AI and agents to roughly 20% of global online orders last Cyber Week, and expects 20% of this season’s ecommerce traffic to come from AI chat agents – but those are influence and traffic measures, counting recommendations, AI-assisted customer service, and referrals, not purchases an agent completed on its own. The distinction will matter when the season’s headlines land.

If no credible measurement of agent-completed orders appears by January 31, we’ll mark this prediction unscored rather than claim a win by default. And measurement itself is part of the story: Meta’s Muse, for example, doesn’t identify itself in a way standard web analytics can see, so some agent activity is invisible to the tools everyone relies on.

A few other things the data shows

Compiling a season’s worth of statistics from more than a hundred sources turns up patterns that don’t fit neatly into a prediction. A few worth knowing as the coverage ramps up:

“Online sales” means at least three different things. For the 2025 season, Census e-commerce for the fourth quarter was $365 billion, Census nonstore retail for November and December was $274 billion, and Adobe’s November–December online figure was about $253 billion. All three measure different things, and seasonal coverage routinely mixes them.

Per-shopper spending surveys can’t be compared with each other. Deloitte’s planned-spend figure is around $1,650 per shopper; Circana’s is around $780. The difference is mostly definitional – what counts as holiday spending, and who counts as a shopper – not a disagreement about behavior.

Mobile wallets keep gaining. Salesforce counted mobile wallets at 29% of US Cyber Week online orders last season, above its own 25% forecast. Checkout friction keeps falling, which matters for impulse-driven categories.

Deal anxiety is real. YouGov finds 36% of US adults are likely Black Friday/Cyber Monday shoppers, rising to about 47% of Gen Z and Millennials, and 42% of shoppers say finding the best deal is genuinely difficult. That anxiety is a big part of why AI research tools are getting traction.

How we are scoring ourselves

All four predictions are logged on the Holiday Shopping Tracker with today’s date, alongside everyone else’s. Here’s the grading schedule:

  • Early January: the BLS December jobs report settles the hiring call, and Adobe’s season recap settles the AI-traffic call.
  • Mid-January: Census December retail sales settle the total-sales call, and December CPI settles the real-growth call.
  • January 31: the cutoff for any credible measurement of agent-completed purchases. If none appears, that half of prediction four goes down as unscored.

We’ll publish the scorecard in January regardless of how it looks. If the consensus turns out closer than we are, that’s worth knowing too, and it’ll be on the tracker either way.

In the meantime, the tracker is free, updated through the season, and built so you can check any number in this post against its source. If you read the data differently than we do, we’d genuinely like to hear it.

The Agile Brand Guide®
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.